TW201447335A - Synchronized pulsed LADA for the simultaneous acquisition of timing diagrams and Laser-induced upsets - Google Patents

Abstract

Method to extract timing diagrams from synchronized single- or two-photon pulsed LADA by spatially positioning the incident laser beam on circuit feature of interest, temporally scanning the arrival time of the laser pulse with respect to the tester clock or the loop length trigger signal, then recording the magnitude and sign of the resulting fail rate signature per laser pulse arrival time. A Single-Photon Laser-Assisted Device Alteration apparatus applies picosecond laser pulses of wavelength having photon energy equal to or greater than the silicon band-gap. A Two-Photon Laser-Assisted Device Alteration apparatus applies femtosecond laser pulses of wavelength having photon energy equal to or greater than half the silicon band-gap at the area of interest. The laser pulses are synchronized with test vectors so that pass/fail ratios can be altered using either the single-photon or the two-photon absorption effect. A sequence of synthetic images with error data illustrates timing sensitive locations.

Description

Simultaneous acquisition of timing diagram and laser induced disturbance, laser assisted device modification system with synchronized pulses System and method

The case claims US Provisional Patent Application No. 61/804,696, application date March 24, 2013, US Provisional Patent Application No. 61/806,803, application date March 29, 2013, and US provisional patent application Case No. 61/838,679, the priority date of the application date of June 24, 2013, the entire disclosure of the above application is incorporated herein by reference.

The present invention is funded by the United States Advanced Intelligence Research Program (IARPA) of the Office of the Director of National Intelligence (ODNI), based on the United States Air Force Research Institute (AFRL) Contract No. FA8650-11-C-7104. The methods and conclusions contained in this case are the property of the inventor and should not be construed as necessarily a guarantee by ODNI, IARPA, AFRL or the US government express or implied.

The technical field of the present invention is a technique for performing defect location analysis by using laser light in integrated circuits (ICs). More specifically, the present invention relates to a technique for performing debug and/or failure analysis of integrated circuit design using Laser-Assisted Device Alteration (LADA).

The principle of the so-called LADA technology is to use a continuous wave (CW) laser to illuminate the back side of the integrated circuit, and generate a local photocurrent inside. (photocurrents), and thus changing the test excitation signal applied to a "suspected" transistor, passing/not passing the test results to locate the suspected defect region, which can contain design or process defects. The laser is used to temporarily change the operational characteristics of the transistor in the device. When a CW laser of 1,064 nm is used, its current spatial resolution is 240 nm.

A description of the LADA technique can be found, for example, in Jeremey A. Rowlette and Travis M. Eiles, Critical Timing Analysis in Microprocessors Using Near-IR Laser Assisted Device Alteration (LADA), published in International Test Conference, IEEE, IEEE Paper 10.4. Pp. 267-279, 2003. This report describes the possibility of using a CW laser with a wavelength of 1,064 nm or 1,340 nm. It is also stated that the 1,340 nm wavelength may cause a change in device operation due to local heating, and the 1,064 nm wavelength may cause a change in operation of the device due to the generation of photocurrent. It should be noted that the 1,064 nm laser has the advantage of spatial resolution. Therefore, the authors recommend using a 1,064 nm laser.

Figure 1 shows a conventional LADA system that uses a continuous wave laser to introduce photonic electron hole pairs from the back side of the wafer into a device under test-DUT. A DUT 110 is coupled to a tester 115, such as an Automated Testing Equipment (ATE), which is coupled to a computer 150. The ATE operates in a conventional manner by issuing a test vector to excite the DUT and analyzing the response of the DUT to the test vector. The ATE can be replaced by a system test board, optionally connected to a computer or similar device for control. Therefore, in this article, "Testing Equipment" (TE) is sometimes referred to as an ATE or other device that tests the DUT. Conversely, when referring to ATE, the reference also refers to the use of other test devices. The DUT is produced by the test vector The raw reaction can be further studied using LADA. For example, if the DUT fails in some tests, LADA can be used to detect if the DUT can pass the test under certain conditions. If so, further determine which of these devices (ie, the transistor) is the cause of the flaw. Conversely, if the DUT passes certain tests, LADA can be used to detect under which specific conditions the DUT cannot pass these tests. If such a situation is found, it can be further verified that the device (ie, the transistor) in it is the main cause of the test failure.

The LADA system shown in FIG. 1 operates in the following manner: The beams generated by the CW laser 120 are concentrated using polarizers 130 and 135 and an objective lens 140, and the DUT 110 is scanned. This operation allows the laser 120 to produce photocarriers in the crucible material. The resulting pair of electron holes affects the timing of the adjacent transistors, that is, shortens or lengthens the switching time of the transistors. The tester is constructed such that the device to be tested applies a selected voltage and frequency in a timed repeating test loop manner, and pushes the operating point of the device to be tested to a critical state. Thereafter, laser excitation is used to change the pass/fail state of the test device test results. The position at which the beam is projected at each point can be correlated with the test pass/fail results produced by the tester. Therefore, when a state change is detected, that is, the transistor that has passed the test before fails to pass, or the transistor that failed the previous test passes, the coordinates projected by the laser beam at the time point indicate that it is "critical". The position of the transistor.

During the LADA analysis, the tester is built to push the operating point of the device under test to a critical state. The way is to use laser excitation to change the pass/fail status of the test results of the test device. The above-described prior art laser assisted testing technique can assist in finding the position of the chirp in space with a resolution of about 240 nm. Further improvements in resolution for single photon LADA are limited by the wavelength of the laser light used. As documented in the Rowlette literature, This spatial resolution can be improved by using shorter wavelengths. However, if a wavelength of less than 1064 nm is used, it will be absorbed by the germanium material, which is the biggest obstacle to providing short-wavelength laser light from the back side to the transistor. Therefore, as recent design rules require a reduction in device size, the spatial resolution provided by conventional LADA systems cannot be improved by using smaller wavelength lasers. For example, under the 22nm design rule, it is doubtful whether a traditional LADA device has the ability to resolve a particular transistor from four adjacent transistors.

Optical beam induced current (OBIC) is another test and debug analysis technique. This technique illuminates the DUT with a laser beam. However, unlike LADA, OBIC is a statistical test method. That is, instead of providing an excitation signal to the DUT, a laser beam is used to generate an induced current in the DUT, and the induced current is measured with a low noise, high gain voltage or current amplifier. OBIC has been used in the single photon mode in the past, and is also used in the two-photon absorption mode, which is commonly referred to as TOBIC or 2P-OBIC (two-photon optical beam induced current).

The two-photo absorption-TPA technique absorbs two photons of the same or different frequencies simultaneously to excite a molecule from a state (usually from the ground state) to an electronic state with a higher energy. The wavelength is chosen such that the sum of the photon energies of the two simultaneously arriving photons is equal to the energy difference between the low energy state and the high energy state of the molecule. The two-photon absorption technique is a second-order procedure whose intensity value is orders of magnitude smaller than the intensity value of a linear (single-photon) absorption technique. The difference from linear absorption is that the intensity of its absorption is proportional to the square of the intensity of the light, and is therefore a nonlinear optical measurement method.

The following summary of the invention is provided as a basis for several aspects and technical features of the invention. This understanding. The invention is not intended to be exhaustive or to limit the scope of the invention. The sole purpose of the invention is to be construed in a single

The various embodiments disclosed by the present invention improve the spatial resolution by using a time domain measurement method, thereby improving the spatial resolution of error localization. Embodiments of the present invention replace pulsed continuous wave lasers with pulsed lasers of sufficient energy. The pulsed laser is synchronized with the clock signal of the device, thereby improving spatial resolution. Various embodiments of the present invention utilize a 1,064 nm wavelength laser to single photon LADA or a longer wavelength to excite a nonlinear two-photon absorption mechanism to produce an inductive LADA effect. The laser technology used in this case can be called a two-photon laser assist device modification (2pLADA) technology.

The disclosed embodiment uses a test vector to excite a DUT while scanning a test area of the DUT using a femtosecond, picosecond, or even nanosecond pulsed laser, and verifying the DUT pair test while scanning. The response of the vector to locate errors within the device. The laser source chosen is one that provides its wavelength to provide photon energy below the bandgap of the germanium material and provides a femtosecond pulse width. The pulse width that provides the best resolution is related to the operating frequency of the device. A clock signal is obtained from the ATE, fed to the DUT and a circuit that controls the pulsed laser. The timing of the pulses can be translated relative to the ATE clock to detect the pass/fail characteristics of each of the devices. In addition, by synchronizing the laser pulses to the clock signal using an appropriate method, the spatial resolution of the measurement can be improved, so that a plurality of devices, that is, transistors, can be identified within the range of the laser beam.

Still further, in accordance with an embodiment of the present disclosure, the laser beam is intermittently parked to illuminate selected areas on the DUT and to collect LADA data at each location. Laser The time at which the pulse reaches each position changes with respect to the clock signal. It is thus possible to construct a profile of the collected LADA data for all illumination locations to investigate the behavior of the device over time, i.e., the response of the device to the relationship of the arrival time of the laser pulse relative to the test vector.

In accordance with a further aspect of the present invention, the LADA test can be applied to investigate the generation of a single event disturbance within a DUT.

Several embodiments of the present invention provide a laser assisted device modification (LADA) system that can be coupled to an automated test equipment (ATE) for detecting integrated circuit devices (DUTs) under test and including: a timing control electronic component Receiving a clock signal from the ATE, the timing control electronic component is configured to generate a synchronization signal for synchronizing the laser pulse to the clock signal; a pulsed laser light source for generating the laser pulse; An optical device for receiving a laser pulse from the fine-tunable pulsed laser source and directing the laser pulse to a desired position on the device under test (DUT); a controller configured to operate the timing control An electronic component, set a time during which the laser pulse reaches the transistor in the DUT, is synchronized with the clock time, and may incorporate a delay or advance of the laser pulse with respect to the clock time, thereby changing the An electrical response of the transistor to the test signal applied to the DUT by the ATE; a single pixel sensor for detecting the reflection of the laser pulse from the DUT and generating a corresponding intensity signal, and wherein the control Constructed to detect the altered electrical response of the transistor, receive the corresponding intensity signal, and utilize the electrical reaction and intensity signal to generate a pattern indicative of an electrical response in a selected region of the DUT Relative to time.

In accordance with several aspects of the present invention, the present invention discloses a method for extracting individual timing diagrams and flip-flops from a synchronous single photon or two-photon pulse LADA. The method can spatially locate (or "stay") the incident laser beam at a target circuit position and detect it one by one according to time. The arrival time of the laser pulse is related to the tester clock signal or the loop length trigger signal, and then the strength and characterization of the failure rate characteristic obtained by each laser pulse arrival time is recorded. The method may include an analysis of a pair of photon-induced disturbance locations to provide a flip-flop transistor perturbation pattern and a logic state distribution pattern, and to graphically display the circuit analysis and image set. The present invention achieves non-destructive, two-photon absorption-induced disturbance location detection, providing an LADA-based implementation that replaces conventional single event perturbation excitation and evaluation methods.

According to another aspect of the present invention, a method for simultaneously obtaining a plurality of timing charts is provided, the method comprising: detecting, by time, one by one, a relationship between an arrival time of the laser pulse and the tester clock signal or the loop length trigger signal At the same time, the LADA activity is extracted from the multiple structures in the defined target area in a LADA image; and the change of the switching event of the transistor on the time axis (ie, the timing diagram) is taken.

According to still another aspect of the present invention, the present invention provides a method for utilizing a server controlled CAD and circuit diagram management device to obtain an extrapolated physical region range within the DUT to synchronize single photon or two photon pulse excitation by LADA The resulting timing diagram. The steps include: a. previewing the location of the CW LADA activity in a controlled target space area; b. recording the specific active location for the CAD set; c. selecting the associated unit block or a larger physical area. d. Construct a logical path between LADA active locations; e. Verify the target region using a synchronized single photon or two-photon pulse LADA; f. Obtain the generated timing diagram; g. Obtain an associated circuit diagram of the associated network; Find the next LADA in the established order; and i. Repeat steps e-h until all components have been verified.

According to a further aspect of the present invention, there is provided a method of displaying a failure rate of a transistor in a DUT, comprising the steps of: receiving test pass and failure data from a tester; receiving a charge injection injected into the DUT. Location information; configuring a display display pixel according to the position on the DUT; storing each test pass result as the first pixel color of the display; storing each test failure result as the second pixel color of the display; and repeating The pixels of the display are examined one by one to accumulate/average the results of the test procedure repeated for a target region in the DUT, and the charge injected for each pixel of the region. If a gray scale display is used, the first color can be white and the second color can be black.

In an embodiment of the present invention, a display method for integrating test success and failure results is disclosed to effectively present grayscale images of black points and white points in a time series. The number of times a pixel stays in a particular spatial location, the attenuation of the phosphor, or the residual image in the retina can effectively average the pass and fail data to provide a color within the grayscale range. When the test procedure is performed, if an image appears in the background gray scale containing 50% black and 50% white during photon injection, it is determined that a target area has been found. Although the observations can also be captured photographicly, the stored data can be stored for non-transitory media for further analysis to obtain the location of the circuit diagram or layout corresponding to the DUT. Using an accumulator circuit to store positive and negative values provides a non-zero number of test results for each pixel in a block. The region can then be enlarged to provide sufficient resolution for analysis of a pattern joint or pattern, revealing the location of the joint portion that produces the transition during the time period to be tested.

Other aspects and features of the present invention will be apparent from the description and appended claims. It must be noted that the detailed description and drawings only provide the present invention. Various non-limiting illustrations of various embodiments. The scope of the invention should be defined by the scope of the appended claims.

1‧‧‧pulse laser source

2‧‧‧Integrated circuit

3‧‧‧Intermediary phase-locked loop circuit

4‧‧‧Laser Scanning Microscope

5‧‧‧Customized application interface

6‧‧‧ computer

7‧‧‧DSP line

8‧‧‧Data Conditioning Circuit

9‧‧‧Oscilloscope

110‧‧‧Device under test

115‧‧‧Tester

130‧‧‧Polar mirror

135‧‧‧ polarizer

140‧‧‧ objective lens

150‧‧‧ computer

210‧‧‧Device under test

215‧‧‧Tester

223‧‧‧Laser pulse sequence

224‧‧‧Laser pulse sequence

225‧‧‧pulse laser source

227‧‧‧Laser pulse sequence

229‧‧‧Laser pulse sequence

230‧‧‧ polarizer

235‧‧‧ polarizer

240‧‧‧ Objective lens

245‧‧‧ signal

250‧‧‧ computer

255‧‧‧Fixed pulsed laser source

260‧‧‧Sequence Control Electronic Components

265‧‧‧Sequence Control Electronic Components

275‧‧‧ phase adjuster

810‧‧‧Device under test

815‧‧‧Tester

823‧‧‧pulse

825‧‧‧pulse laser source

830‧‧‧ polarizer

835‧‧‧ polarizer

837‧‧‧half mirror

840‧‧‧ Objective lens

842‧‧‧focus lens

845‧‧‧ signal

850‧‧‧ computer

860‧‧‧Sequence Control Electronic Components

870‧‧‧ optical sensor

872‧‧‧ signal

900‧‧‧Test system

905‧‧‧Tester

910‧‧‧microchip

920‧‧‧clock and pulse controller (schedule)

928‧‧‧Laser pulse

930‧‧‧Electrical charge injector (injector)

940‧‧‧Device under test

960‧‧‧Location pass/fail data storage (accumulator)

The accompanying drawings are included in this specification and form part of this specification. The drawings illustrate embodiments of the invention and together with the description The figures are intended to describe the main features of the illustrated embodiments in the form of a chart. The figures are not intended to describe each feature of the actual embodiment or the relative size of the components shown in the description, and are not to scale.

1 is a system diagram depicting a known technology CW LADA system.

2 is a system diagram depicting a pulsed laser LADA system in accordance with an embodiment of the present invention.

2A is a schematic diagram depicting an embodiment of two feedback loops of the present invention.

2B is a schematic diagram depicting the use of a fixed pulsed laser source to generate a clock signal in accordance with an embodiment of the present invention.

3 is a block diagram depicting an embodiment of the synchronization mechanism of the present invention.

4A-4C are schematic diagrams showing how the pulsed laser LADA of the present invention recognizes and separates individual transistors in adjacent PMOS and NMOS transistors.

5A-5D are schematic diagrams illustrating a method for improving spatial resolution by providing accurate positioning of laser pulses.

6 is a schematic diagram depicting a laser repetition rate locking mechanism in accordance with an embodiment of the present invention.

7A-7C are schematic diagrams depicting the steps of measuring a minority carrier life cycle using LADA, in accordance with an embodiment of the present invention.

8 is a system diagram depicting an embodiment of a pulsed laser LADA system of the present invention, FIG. 8A showing a graph generated using the system of FIG. 8, and FIG. 8B showing a graph of image change and time versus time.

9 is a block diagram showing the architecture of a pulsed laser LADA system in accordance with an embodiment of the present invention, which system can be used in the system of FIG.

Figure 2 shows an embodiment of the invention. The system in the figure uses a pulsed laser source that produces enough energy instead of a continuous wave laser. The technique of this embodiment uses the photon absorption principle to accurately inject a carrier into an integrated circuit (IC) for judging the location of the error in the device under test using the LADA technique, and can be used for characteristic analysis of the IC and to seek improvement. The method of IC design. The main principle of this technique is to deliver photons to the focal point on the transistor such that the transmitted photon energy is greater than the energy required to generate the electron hole pair. For example, if it is enamel, it is greater than 1.1 electron volts (eV). For other materials such as gallium arsenide (GaAs), germanium (SiGe), indium phosphide (InP), etc., they have different energy band gap energies. In this embodiment, photon stimulation needs to be excited by a laser pulse in the nanosecond to femtosecond range. This signal is concentrated at the focus of the laser, providing an immediate improvement in the location of the error location. Due to the control of synchronization, the effective number of electron hole pairs appears to be reduced. However, this embodiment utilizes sophisticated timing control electronics to precisely control the timing of the laser pulses to match the transition edges of the test device's clock signal (eg, ATE clock signal). This type of control can be used to make slight changes in the delay or advancement of the signal as it propagates in a suspected transistor in the Laser-Assisted Device Alteration (LADA).

Figure 2 shows an embodiment of the invention. The figure shows a DUT 210 coupled to a ATE 215, the same as the prior art. However, in the embodiment of FIG. 2, nanosecond to femtosecond pulsed lasers are generated by pulsed laser source 225 and then concentrated onto polarizer 230 and 235 and objective lens 240 onto the DUT 210. In the 2pLADA of the present invention, the laser source 225 provides a pulsed laser beam having a wavelength band longer than 矽, that is, a pulsed laser beam having a wavelength exceeding 1,107 nm. In one embodiment, a wavelength of 1,550 nm is used, while in other embodiments a wavelength of 1,340 nm or 1,250 nm is used. On the other hand, the same device can also be applied to single photon LADA. In this case, the laser source can provide a pulse beam wavelength of, for example, about 1,064 nm. In this embodiment, the polarizers 230 and 235 are implemented by a Laser Scanning Microscope (LSM). Also, in some embodiments, a Solid Immersion Lens (SIL) is used as part of the configuration of the objective lens 240.

In a conventional LADA system, the laser source is normally open (ON). However, in the embodiment of the present invention, it is a pulse that is extremely short in use. Therefore, it is extremely important to allow the device to change state when the laser pulse arrives. To achieve this, the present invention uses a trigger signal 245 that is taken from the ATE and input to timing control electronics 260 to control pulsed laser 225 such that its laser pulse is synchronized with the test signal of the ATE.

When using the system shown in Figure 2, the test equipment (ATE) 215 is first operated to determine the critical setpoint of the DUT 210 from a set of test vectors. That is to say, the voltage and frequency of the test vector are changed at any time to judge whether the DUT is close to the condition that cannot pass the test, or is reaching the critical point when the test fails. For example, the number of failures of the DUT in the test reaches a test cycle. 50%. This is the test of the DUT passing/not passing the critical point condition. The set of voltage and frequency at this time will then be used to generate a repeated test signal to repeatedly excite the DUT under the pass/fail conditions of the test.

While the DUT is being excited under its critical conditions, the tester 215 sends a synchronization signal 245 to the timing control electronics 260. The timing control electronics 260 controls the laser source 225 to produce a laser having a picosecond to femtosecond pulse width and a wavelength greater than the chirp bandgap (for 2pLADA) or a laser shorter than the chirp bandgap Pulse (for single photon LADA). Generally, the wavelength for 2pLADA is between about 1,250 nm and 1,550 nm, and the pulse width is about 100 fs. The single photon LADA has a wavelength of 1,064 nm and its pulse width can be in the range of nanoseconds or femtoseconds. The laser pulse scans a region to be tested of the DUT 240 to extend or shorten the switching time of the DUT and force the DUT beyond the critical point. This is done by adjusting the timing of the laser pulse to a value that would cause the DUT test to fail if the voltage/frequency of the test vector is set such that the DUT is close to a value that cannot pass the test. Conversely, if the voltage/frequency of the test vector is set such that the DUT is reaching a test failure value, the timing of the laser pulse is adjusted to a value that will pass the DUT test. During this time, the output of the DUT is monitored to determine where the error occurred in the device. The method is that the signal outputted by the DUT indicates that the device cannot pass the test (if the laser beam does not exist, the DUT can pass the test), and the position of the beam projected on the DUT is determined, and the position is As the location of the transistor that caused the error. Conversely, the signal display device outputted by the DUT passes the test (if the laser beam does not exist, the DUT cannot pass the test), and determines the position of the beam projected on the DUT, and uses the position as the previous error. But now it passes the position of the tested transistor.

It must be noted that since the test device generates a synchronization signal, the timing of the laser pulse can be changed to change the amount of action of the optically generated (single-photon or two-photon) effect on the transistor. The way is to change the timing of the laser pulse to extend or shorten the DUT Switching time. In addition to being used to determine the location of the error, this test function can determine the severity of the error.

Embodiments of the present invention also use timing control electronics to precisely control the timing of the point at which the laser pulse rises and falls relative to the clock signal of the test device (e.g., ATE). In this way, the delay can be fine-tuned or the propagation of the signal within the target transistor can be accelerated. In one embodiment, as depicted in Figure 2A, two Phase Locked Loops (PLLs) are utilized to accurately control the pulsed laser. In FIG. 2A, the ATE 215 provides a clock signal Clk and a test loop signal Test Loop. The clock signal and the test loop signal are both input to the DUT and simultaneously transmitted to the timing control electronics 260 to form a first PLL. The laser source 225 then includes a second PLL.

That is, the PLL of the laser source 225 ensures that the pulse frequency of the laser pulse is stable and accurately reaches a desired frequency, for example, 100 MHz. In contrast, the first PLL of the timing control electronics enables the second PLL frequency to be synchronized to the clock signal of the ATE. It should be noted that the term "synchronization" as used in the present invention does not necessarily mean that the laser pulse coincides with the clock pulse, and it can also mean that the two can form a synchronization during a test loop. Thus, for example, the timing of the laser pulses can be translated such that the pulses are generated in the middle of each clock pulse of the clock signal, as shown by pulse sequence 227 in the figure. It can also be generated at the end of each clock pulse, as shown by pulse train 229. The rest of the way. That is, the laser pulse may be delayed or earlier relative to the clock signal of the ATE, but remains synchronized to the clock signal of the ATE.

Alternatively, the frequency of the laser pulse is set to a multiple of the ATE clock signal as described below. For example, the number of pulses of the laser pulse sequence 223 is seven times, and seven laser pulses are generated during each clock pulse of the ATE. Detecting device errors with a multiplier greater than one It is produced at rising edge, falling edge or other time. At the same time, since several laser pulses are provided with an early or delayed display function with respect to each clock pulse, it is not necessary to delay or translate the pulses. Conversely, a multiplier less than one can also be used. For example, in pulse sequence 224, the multiplier is 0.5 times, so that every other clock signal has a laser pulse arriving. This design can be used to verify that the device's defect is indeed due to the laser pulse, because if the cause is from the laser pulse, the device will fail to detect 50%.

Figure 3 shows a mechanism for achieving synchronization in an embodiment of the invention. It can be output from a pulsed laser source 1 via a medium phase locked loop (PLL) circuit 3, which is a nanosecond to femtosecond pulse, synchronized to the clock period of an integrated circuit (IC) 2. Under this design, the PLL circuit accepts the clock cycle frequency of the IC and locks it to an internal crystal oscillator having the same oscillation frequency as the clock signal. In this embodiment, the clock frequency and the frequency of the quartz oscillator are both fixed at 100 MHz, and the IC clock signal can be generated by the ATE (not shown). This design is capable of forming a 1:1 ratio of light pulse to transistor switching event synchronization. Under this condition, in practice, the frequency can be set to any value between 1 kHz and 10 GHz, and then eliminated one by one according to the efficiency of the photon absorption rate to obtain a desired value.

It should be noted here that sources above 1 GHz are not suitable for non-linear measurements, such as the 2 pLADA method of the present invention. This is because the peak optical power of each pulse is usually inversely proportional to the repetition rate. Therefore, the high repetition rate is equal to the low peak optical power, and the resulting two-photon absorption efficiency is not sufficient even if it exists. Conversely, if a single photon measurement using 1064 nm for 1:1 synchronization is used, a light source of several GHz is advantageous because its photoelectric effect is scaled with the amount of energy of the incident light. In addition, it should also be noted that the efficiency of two-photon absorption is directly proportional to the period of the incident pulse. Therefore, the femtosecond light pulse produces a peak light energy that is more than picosecond or nanosecond. It is much higher, so it can improve the nonlinear absorption phenomenon. Therefore, in non-linear measurement applications, ultra-high speed (femtosecond or picosecond) light pulses must be used. On the other hand, in the single photon measurement application, the pulse period is not a limiting parameter for the absorption rate, so it does not affect the measurement effect. Rather, additional probing parameters can be provided, such as can be used to measure the duration of the interaction of the optical pulse to the optoelectronic device excitation. Not only that, the absorption coefficient of 矽 for single photon wavelength (less than 1,130 nm) is larger than that of two-photon absorption fine-tuning (ie, >1250 nm).

In order to maintain the efficiency of the system, the synchronization mechanism can be modified to match an integer multiple of the incident light pulse to the switching event of the transistor (or the clock frequency of the device). To achieve this, the laser source is designed to produce a repetition rate greater than 1 GHz, and has a scalable pulse selection module that is placed after pulse optimization to modify its synchronization ratio. For example, each of the incident light pulses may not be matched to the switching events of the respective transistors, and each of the second pulse formations may be matched to each of the next switching events, thereby producing a 2:1 synchronization ratio. In practical applications, this can be achieved using a 200MHz repetition rate laser and a 100MHz device frequency, or a 1GHz repetition rate laser and a 500MHz device frequency. The rest of the way. Alternatively, the ratio can be adjusted to 3:1 or 4:1, etc., as long as the ratio corresponds to a photo-electrical timing such that its clock pulse can be synchronized with the test loop signal. Under this synchronization mechanism, the efficiency of photon absorption does not decrease, but the ratio of absorption occurs, so that the intensity of the generated photon signal will form a negative scaling ratio with its synchronization ratio. It should be noted that the above is not a limiting parameter for laser-induced measurement of integrated circuit technology. For each device under test, photon scaling correction must be performed to determine the maximum allowable value of the synchronization ratio. In addition, if a fine-tunable light source (ie, 1000-1600 nm output wavelength) is integrated into such a system, it can be applied in both single photon absorption and two-photon absorption mechanisms. exchange. Because when the wavelength is greater than about 1200 nm, two-photon absorption will be significantly greater than single photon absorption.

Once the above frequency (ie, the clock frequency and the frequency of the quartz crystal oscillator) is locked together, the output signal of the PLL circuit is transmitted to the pulsed laser via an electronic filter of 100 Mz (or the frequency of the clock). Act as an excitation signal for its input. The advantage of this approach is that the PLL line has full control over the phase of its output signal. Therefore, it can be used to control the repetition rate of the laser light output, thereby controlling the arrival time of its pulse. This can be verified by the oscilloscope 9 comparing the clock frequency output from the IC with the excitation output output from the pulse source. In this example, the PLL line can electronically achieve a phase delay of approximately 600 fs; however, since the board is electrically dithered, the minimum phase delay is set to approximately 20 ps. The electrical jitter of the system is directly proportional to the accuracy of the optical pulse relative to the location of the transistor switching time in the individual verification. Therefore, from the electrical jitter of the system is 20 ps, the optical position is also accurate to 20 ps, forming a one-to-one pairing. This is a critical parameter because if the optical position error is greater than, for example, the femtosecond pulse period of the 2pLADA, the efficiency obtained by the above timing control will be offset. Femtosecond light pulses can increase the energy density of the region to achieve effective two-photon absorption, but when the electrical jitter erodes the time of generation of the isolated carrier, the jitter will limit the generation of the next signal, and at that time The time accuracy of the time solution data.

The laser pulse is then coupled to a Laser Scanning Microscope (LSM) 4 for accurately distributing the pulse at a particular location on the IC. The LSM is controlled by a computer 6 that provides a graphical user interface and a custom Digital Signal Processor (DSP) package. In the disclosed embodiment, the application package provides a function for the end user to directly exchange data with the PLL line via a predetermined DSP line 7, and the PLL line provides time for the laser pulse to arrive at the device. Full control For example, the pulse can be delayed or advanced.

The device 2 will now be described. The device 2 can generate a LADA pass/fail value of a preset condition due to electrical excitation, and is displayed in a customized application interface 5. The interface panel compares the instantaneously obtained value obtained from a combination of a counter, a latcher and a displacement register with a preset reference value inserted in advance to select a reset switch. The precise control of the instant set counter value can be achieved by an analogy trimming delay potentiometer. The analog-type delay potentiometer is provided on the application panel to change the timing of the function of the clamp to the IC. The above architecture provides the user with conditional control such that the output value of the comparator becomes pass, fail or half. The pass/fail output values described above are then provided to a data conditioning circuit (in this example, a utility programmable gate array FPGA)8. The circuit is programmed to receive an immediate digit pass/fail signal to conditionally convert the failed (not passed) value to a value of 0-100% and output an average (with a period of approximately 40 us) The digital output is also represented by a 0-100% failure value to improve its visibility, and the resulting pass/fail degree is estimated and displayed on the graphical user interface. The data conditioning circuit can also be used to interface with the application panel to calculate the magnitude of the laser induced timing delay after correcting the voltage output by the application panel.

In the above embodiment, a finely tunable pulsed laser source is used and the pulse frequency is adjusted to synchronize to the ATE clock. While the above embodiments are feasible, the finely tuned pulsed laser source is typically extremely expensive and requires the use of the PLL described above. Figure 2B shows another embodiment of the invention in which LADA measurements are achieved using a simplified fixed pulse laser 255. For example, a modal locked laser source can be used. Modal locking refers to an optical technique that can be used to generate very short laser pulses during a period of picosecond or femtosecond. The laser pulse can be applied as a clock signal to the timing control electronics 265. Traditional ATE has a clock input port, so it can be programmed The input clock signal is programmed to generate a clock signal Clk for use by the DUT, as well as a test loop signal. Thus, in one embodiment of the invention, the clock signal of the timing control electronics 265 is input to the ATE, and the ATE is programmed to generate a clock signal and a test loop signal using the input signal.

However, as described above, in order to apply the pulsed laser LADA to the extreme, its pulse must be adjusted such that its laser pulse reaches the transistor at different points in the clock cycle, for example, at the rising edge of the clock cycle, The intermediate point or the falling edge is equal to the time point and reaches the transistor. In the embodiment shown in Figures 2, 2A and 3, the laser pulse is achieved either early or delayed. However, in the embodiment of Fig. 2B, the laser pulse is fixed and cannot be changed, and therefore cannot be implemented in a manner that delays or advances the laser pulse. To this end, in one embodiment of the invention, the ATE is programmed to delay or advance its clock signal to synchronize with the clock signal derived from the timing control electronics 265. In this way, the point at which the laser pulse reaches the transistor can be fine-tuned to the rising edge, the falling edge, and the like of the ATE clock signal.

On the other hand, because the ATE and the LADA tester are usually manufactured by different manufacturers, and the test is actually performed by a test engineer of another third company, the operation of the test engineer can be simplified, and the ATE delay is eliminated. Or the task of the signal early will be more advantageous. To achieve this, the phase adjuster 275 shown in the embodiment of Figure 2B can be used. This is accomplished by using the phase adjuster 275 to cause the clock signal output from the timing control electronics 265 to be earlier or later than the laser pulse. The resulting adjusted signal is then sent to the ATE as an input clock signal. In this way, when the ATE outputs its clock signal and test loop signal, both signals can be translated or delayed relative to the laser pulse.

Example

A pulsed LADA system with a pulsed source is constructed to provide a new orientation for evaluating or measuring the operating device. Traditional single-photon or alternative two-photon LADA uses a CW laser, and the optical radiation used continues to interact with individual transistors, making it invasive to the extent that it can be compromised. On the contrary, the present invention uses a pulse type LADA technology, which enables the switching behavior characteristics of individual transistors to be distinguished in two physical dimensions. The extended concept of this pulse type LADA will be discussed below.

Under the traditional CW LADA excitation, the theory and practice of the device have confirmed that the laser-induced device perturbations from a p-type metal oxide semiconductor (PMOS) transistor will exceed its adjacent n-type metal. The strength of an oxide semiconductor (NMOS) transistor. However, since the diameter of the laser beam may cover the p-type and its adjacent n-type semiconductor, the resulting spatial resolution is still insufficient to distinguish which is a germanium transistor. On the other hand, if the embodiment of the present invention is used in the form of a pulse, the spatial resolution can be improved by the temporal resolution; even the laser with a longer wavelength can be used to obtain the same result. In other words, since the incident pulse has been fine-tuned to be exactly the same as the switching time interval of the transistor under verification, and since the peak energy contained in each pulse is much larger than in the case of using the CW technique, it can be from the transistor in the immediately adjacent position. Differentiate and separate individual PMOS and NMOS transistors. This was not possible with the CW excitation method in the past. The present invention has devised a new experimental method for the design and debugging of semiconductor devices, and can be applied to the design rules of increasingly small and small. The method of the present invention has solved a major problem in the semiconductor device failure analysis industry. This problem stems from the fact that the recent technology nodes have been reduced to lower nanogeometry conditions, making optically induced transistor identification and operational characteristics analysis more important, but there is no solution. Therefore, the synchronized pulse LADA method of the present invention has a higher level than the conventional CW LADA side. The value of the law.

4A-4C show an example of the mechanism of the above improvement. In the continuous wave method, since the PMOS signal usually has a dominant position, only a substantially spatial distribution of a single signal can be obtained, as shown in FIG. 4A. In this way it is extremely difficult to separate the actual distribution position of the individual transistors and/or apply the resulting LADA display results to a computer aided design (CAD) layout. In theory each transistor will produce its own LADA signal, regardless of its laser-induced effect intensity, as shown in Figure 4B. These signals can be used to perfectly track the actual position of individual transistors for fast physical and/or optoelectronic identification. This phenomenon can also be applied to the pulse field and is implemented in the above embodiment. The method is to control the timing of the laser pulse to synchronize with the test signal, and to reach the position of each PMOS and NMOS transistor according to the user's choice. The pulse can be adjusted to coincide with the switching action of the PMOS transistor to test the PMOS transistor, and can also be adjusted to coincide with the switching action of the NMOS to test the NMOS transistor, as shown in FIG. 4C. Therefore, the switching action of the specific transistor can be improved, and the CAD reinforcement entity correspondence and recognition of the LADA signal can be improved, and is not restricted by the spatial coverage of the laser beam.

In addition, the increased peak power produced by the ultra-high-speed pulse can be increased or decreased in addition to the function of the LADA signal, that is, the result is less than the average value of the image (depending on the desired generation of the PMOS or PMOS) Depending on the NMOS transistor, the laser induces critical timing path perturbations, thus improving the collection of LADA signals. The stronger incident light power can increase the number of light-injected carriers in the germanium material and increase the probability of exciting optoelectronic fluctuations in the structure of the device under test. Significant LADA signal responses can be achieved in this manner, but can be measured using only a lower degree of intrusive or energy. The pulse light source is actually turned OFF for longer than the ON time, so heat accumulation and damage are reduced. The chance. For example, with an ultra-high speed laser with a pulse period of 10 ps, the laser is turned off for 10 ns at a repetition rate of 100 MHz, resulting in a 1:1000 ON/OFF ratio, thus providing sufficient cooling downtime. It should be noted, however, that the ultimate cause of heating is its power ratio. For example, if a single light pulse contains 1kJ of incident optical energy, the above conditions can be met, but at the same time, it also contains enough energy and may cause permanent damage to the device due to other thermal or non-thermal photoelectric mechanisms.

At the same time, due to the need to inject a relatively large amount of optical power into a particular transistor in a non-invasive manner, it is of course possible to create a perturbation of the previously neglected transistor position. Originally, it was necessary to generate large-scale photocarriers in the vicinity of a suspected sputum area (which is common for detection of transistors with different sensitivities), which would increase the possibility of improperly expanding the visual range of the LADA detection area. Sex. The region to be activated can be excited using a laser induced photocurrent of about 10-100 uA. However, the peak optical power driven by the ultra-high speed laser pulse is as close as 10-100 kW, which is enough to inject a photocurrent of 10-100 mA into the device to be tested, while still maintaining intrusion in a safe degree. This energy is enough to cause perturbations in "healthy" transistors.

To achieve effective two-photon absorption in the tantalum material, a focal laser power density of more than 10 MW/cm ² (million watts/cm 2 ) can be used. However, when used for single photon absorption, its value is approximately 106 times smaller. This is because its relative absorption cross section is different. To achieve efficient and non-invasive photon implantation, the incident light power (or local power density) needs to be reduced because the geometry of the transistor to be verified has shrunk. At the same time, the occurrence of two-photon absorption does not depend on a specific power density threshold. Two-photon absorption is an instantaneous, quantum mechanically defined nonlinear process that varies with the imaginary part of its third-order nonlinear susceptibility, that is, with the square of the intensity, rather than with the power density. .

Even though the two-photon wavelength of 1,250 nm is effective to produce an absorption of 625 nm inside the crucible (where the absorption cross section is greater than 1,064 nm), the absorption process depends on the strength characteristics and also reduces the total relative proportion of its absorption. Two-photon absorption is directly proportional to the square of the incident light intensity. In addition, the degree of doping of bismuth also affects the result, that is, increasing or decreasing the doping concentration will affect the efficiency of absorption, which is a function of wavelength. However, this single photon-affected probability can be used to implement another novel laser detection and device characterization platform that provides high-precision timing analysis of signal differentiation and conversion levels within the transistor. The traditional CW LADA method does not provide this form of inspection analysis because of its invasive nature (because its laser is normally open) and limited power transfer capability. In contrast, the present invention uses a time-resolved pulse detection method to provide an unprecedented analysis of the flawless, design-defined nodes and their performance and interaction with subsequent downstream devices to verify their transistor switches. The physical characteristics of the action. In order to be able to effectively implement the device characteristic analysis as in the present invention, it is necessary to first understand the required incident light power intensity. To cause perturbation of "healthy" transistors, high peak power is required, but at the same time minimal intrusion must be maintained. Under these conditions, it is extremely necessary to optimize the period of the incident light pulse. As is known, at a wavelength of 1064 nm, a picosecond pulse can provide a comparable intensity of incident light power at the transistor level (and generate enough photocarriers) because, for example, a 10 ps laser pulse, the repetition rate is 100 MHz, When the average power is 4mW, a peak power of 4W can be generated. However, this condition cannot be achieved when the laser repetition rate is matched to a clock frequency higher than 1 GHz obtained from the device under test. Increasing the repetition rate results in a drop in peak power. Therefore, another more suitable alternative is to use a femtosecond laser source. The laser repetition rate can be scaled according to the operating frequency of the device, while providing an increased level of peak optical power, since the pulse period has been shortened by 1000 times, so that the peak power can be increased by the same factor. number. In the above example, it is 4 kW. Another advantage of using a femtosecond pulse period is to improve its time characteristic analysis effect, but the improved effect is limited by the intensity of the electrical jitter of the synchronization mechanism, as explained above. Finally, femtosecond pulses provide a lesser degree of optical intrusion than picosecond or nanosecond pulses, minimizing the likelihood of laser induced damage to the device.

Furthermore, the pulse type LADA system of the present invention has been shown to improve the spatial resolution of detection due to its ability to properly control pulse timing. Compared with the conventional CW method, the CW method is to stimulate the LADA information in real time by exciting a specific area of the suspected sputum by the laser. The result is a spatially average two-dimensional LADA image, because there is no difference between the higher-level order of circuit functionality, ie, the propagation signal path and time; only the combination of the two can be obtained. Distribution information. Among them, due to the PMOS dominance, there is a deviation. In contrast, the pulse type method of the present invention can distinguish paths of different propagation speeds with an accuracy of 20 ps, so that it can be used for high-resolution LADA signal display, and provides improved lateral resolution. The reason is that the results obtained can individually distinguish spatially separated adjacent transistors by time; these transistors are not set to perform switching operations at this time, but are switched on during subsequent device operation cycles. The invention can improve the difference resolution of LADA and the actual LADA resolution.

5A-5D show an example of the architecture of the present invention. In continuous wave mode, since the spatial distribution of the LADA signal is time averaged, the resulting two-dimensional LADA profile can only provide a rough photoelectric structure, and the display result is about the LADA signal strength of the individual transistor (because PMOS is usually Deviation is more dominant than NMOS). The resulting image is shown in Figure 5A, which shows poor spatial resolution and limited ability to CAD drawings. On the other hand, in the pulse mode of the present invention, the obtained LADA image quality is better, and the spatial resolution thereof is improved. Due to the time resolution characteristics of the signal acquisition method. Since the individual locations of individual transistors are set as a function of space and time (and because sufficient incident optical power can be provided to remove possible PMOS bias effects), the effects of adjacent transistors have been obtained by perturbing the LADA signal. Effective removal. A tester drive is obtained that depends on the sequence of events of the transistor that control the operation of the device. Each transistor is set to perform a switching action in a systematic, time-dependent sequence, allowing the incident light pulse to directly depict and measure each transistor in two physical dimensions (ie, X and Y) and on the time axis. . As a result, the spatial resolution of the resulting LADA signal is improved, and thus additional device-related data not available in the prior art can be extracted, as shown in the order of Figures 5B and 5C. The images shown at different points in time show the difference in time and space.

In addition to the ability to collect LADA related data using the present invention, the present invention can also be used to determine other photoelectric phenomena. One example is to measure the life cycle of a laser-induced carrier. Under the prior art, quantification of the carrier life cycle at a particular location on the device is extremely difficult. This is because such measurements require most different optoelectronic parameters, such as material composition, size, geometry, and electric field strength and direction. However, if the pulse type LADA technique of the present invention is used, the electronic time series can be directly measured by a pseudo pump-probe type technique. The measurement is to connect the occurrence of a LADA event of a particular transistor to the arrival time of a laser pulse. The measured carrier life cycle may need to be adjusted (ie, subtracted) by the electronic reaction time of the system to obtain a more accurate measurement.

When using single-photon LADA, a laser pulse with a wavelength of 1064 nm, the intensity of the measured LADA effect is directly proportional to the intensity of the laser-induced photocurrent (this is when the linear absorption is used, the LADA signal is double-photon) Second response). According to an embodiment, the LADA The signal can be compared to a function of the arrival time of a laser pulse. It is then possible to capture the carrier lifetime as this period will indicate the resulting LADA signal strength.

According to an embodiment of the invention, the implementation procedure is as follows. In the first step, a laser beam (e.g., a CW laser beam having a wavelength of 1064 nm) is parked to illuminate a suspected germanium transistor to obtain an optimized LADA signal. As shown in Figure 7A. The spatial coordinates of the laser beam at the best LADA signal, indicating the appropriate spatial coordinates of the transistor. Turning off the CW laser source and turning on the pulsed laser source, directing its laser pulse to the spatial coordinates obtained from the CW laser. The laser pulse time is adjusted to generate and measure the optimized LADA signal. A suitable temporal overlap is obtained with a tester (e.g., ATE) pulse that arrives at the transistor, as shown in Figure 7B. At this time, the spatial overlap of the laser spots on the transistor and the time overlap of the laser pulses on the test signal will reach an optimum value. Moreover, the laser pulse arrival time can be adjusted to match the measurement of the life cycle of the carrier. In particular, changing the timing of the laser pulses, the intensity of the LADA signal at each point in time (eg, a delay or an early amount) is recorded until the LADA signal becomes zero. The plot shows the response of the LADA signal intensity to time, as shown in Figure 7C. The time required for the LADA signal to transition from a maximum value to a minimum value (or vice versa) corresponds to the measured laser induced carrier life cycle. When performing the above procedure, it is necessary to repeatedly apply the electronic test signal to the DUT.

Laser source

The prior art has been able to provide laser sources with repetition rates up to several GHz. The laser light source is precisely regulated to the length of its resonant cavity, that is, the shorter the oscillation cavity, the higher the repetition rate. Control of the length of the chamber can be accomplished by a piezoelectric actuator disposed on the opposite side of the resonator mirror in the cavity. This is the current industry standard repetition rate locking technique, but the electronic mixer circuit used to implement the mechanism of the present invention may differ in design and implementation. To be able to The pulsed laser source is incorporated into the LADA tester of the above embodiment, and two sets of feedback loops are used. One set is used to control the repetition rate of the laser pulse, and the other is used to synchronize the timing of the pulse to the clock of the DUT. The first set of feedback loops for controlling the repetition rate includes a mixer for comparing the free-running repetition rate frequency of the laser with an input clock excitation signal to generate a high voltage driven differential signal. The differential signal is input to the piezoelectric transducer to adjust the length of the resonant cavity, thereby adjusting the desired length to match the repetition rate of the pulse to the input clock signal. An example of this processing mechanism is shown in Figure 6. In addition to the circuitry shown in Figure 6, another secondary stabilization mechanism can be included in the system to continuously monitor and correct the output voltage of the fractional-integer amplifier. In this way it is ensured that the high voltage amplifier can always get the correct input voltage for a stable lock repetition rate over a longer period of time, for example over several days instead of tens of minutes.

Embodiments of the present invention also provide a method for extracting individual timing diagrams from synchronized single photon or two photon pulses LADA. The method includes spatially locating (or "parking" the incident laser beam to a target circuit feature. Then, the arrival time of the laser pulse is checked one by one by time. This time is the time relative to the test clock signal or the loop length trigger signal. The intensity and signal of the failure rate characteristic obtained for each laser pulse arrival time is then recorded. The LADA measurement method can further capture the flip-flop response (ie, logic state control). Traditionally, a single event disturbance corresponds to a particular location on the DUT in a manner that directly measures the intensity of the laser induced photocurrent. The method of the disclosed embodiment of the present invention captures the disturbance information when performing an LADA-based test. The present invention derives a new approach to assessing these perturbations and is more efficient.

The present invention uses other single photon absorption methods, in one embodiment a two-photon absorption method that excites p-junctions and n-junction semiconductors for devices suspected of having time-dependent failures, Provides digital visibility. When the two-photon absorption method is used, a local axial space effect value can be measured for the region where the photocurrent is injected. With a sampling accuracy of 10 picoseconds, a particular test vector edge can be made to be at a critical point of pass/fail in statistical methods.

LADA waveform

Figure 8 depicts an embodiment of an LADA system of the present invention. The system can take individual LADA waveforms at each specific (xi, yi) location. This waveform represents the relationship between the failure rate and the arrival time of the laser pulse. This waveform can help analyze the behavior and/or failure mechanisms of individual transistors in the DUT. The system itself incorporates most of the components of the system of Figure 2, and components similar to those shown in Figure 2 are labeled with similar component numbers, but all begin with an 8 word.

In the system of FIG. 8, an optical element, such as a half mirror 837, is used to direct light reflected from the DUT onto the optical sensor 870. If desired, a focusing lens 842 can be inserted into the path of the light to focus the reflected light onto the optical sensor 870. For example, the optical sensor may be a photodiode, an avalanche photodiode (APD), or a photomultiplier tube or the like. In this embodiment, the optical sensor 870 is a single pixel sensor. When the LSM directs a laser pulse to a selected location of the DUT, the light reflected from this location is induced by the optical sensor 870, so the inductor 870 transmits a corresponding intensity signal 872 to the controller 850. Accordingly, an intensity signal corresponding to the individual illumination locations can be recorded on the controller 850. If the LSM is used to move the laser pulse one by one through adjacent points, so that the laser pulse illuminates a given area on the DUT one by one, the image composed of the intensity signal should be under no other action. It is a fairly consistent gray image, sometimes called a pepper salt image (generated using grayscale image digitization techniques).

In operation, select a specific area on the DUT for verification and test the tester. The test vector is set to produce a 50% failure rate for the device to exhibit within the selected area. Thereafter, the test loop is repeatedly executed so that each pixel is irradiated several times, and the intensity of the reflected light is recorded. The term "pixel" as used in this context refers to a spatial position, that is, the position at which the laser pulse is illuminated when the LSM holds the mirror to a specific position. The intensity signal produced by the optical sensor 870 is considered to be a single pixel. However, when the LSM moves the mirror to the following position on the DUT to cover all of the selected area, the light collected from the optical sensors at each location is considered to correspond to the pixel at that location. In this way, the selected area can be spatially divided into image pixels, although the optical sensor is only a single pixel sensor. Conversely, images produced by sensors located at various locations may be represented by a majority of pixels on the display monitor. The ratio between them depends on the ratio of the diameter of the sensor to the size displayed on the screen when imaging the DUT. In this paper, the area that the sensor images the DUT is called an image pixel, but when displayed on the monitor screen, it may include a large number of screen pixels.

As shown in FIG. 2A, the present invention can control the synchronization of the laser pulses such that the laser pulses arrive at the DUT at a predetermined delay or early time relative to the test vector. In one embodiment, the present invention utilizes such control to set the timing control electronics such that the laser produces a sequence of optical pulses p across the entire test vector signal v, as indicated by the dashed box in FIG. The pulse 823 in each pulse sequence corresponds to a time point ti, and the reflected light signal 872 collected by the optical sensor 870 at each time point ti is recorded. This step is repeated for each pixel so that the system can record the individual bits in the selected area, i.e., the intensity signals at each time point ti at each illumination position (Xi, Yi). If all of the transistors in the selected area respond individually to the test vector and the laser pulse, the resulting image will be a salt and pepper-like image.

However, if the test vector is set to maintain the failure rate of the transistor at At 50%, the laser pulse will force the partial transistor failure rate to increase and the partial transistor pass rate to increase. The inventors have discovered that the pass rate and failure rate of each of the transistors is dependent on the arrival time of the laser signal relative to the clock signal of the test vector. Accordingly, the system of Figure 8 is configured to capture the information and present the information to the user. In particular, in generating a combination of individual pixels and pulse arrival times, for example: (xi, yi, ti), the test is repeated a plurality of times to generate a failure rate value (%). The inventors have also discovered that some of the transistors will increase their pass rate due to the laser pulse, but only the initial set value whose failure rate is less than 50% is recorded. A graph is then generated for the individual locations (xi, yi) to show the relationship between the failure rate change and the time ti, as shown by the solid squares in FIG. The figure shows the relationship between failure rate and time at the selected point (xi, yi) on the DUT.

It should be noted that although the graph is simply referred to as "representing the relationship between failure rate and time", the term "time" refers to the time difference, that is, the time difference measurement between the test signal and the laser pulse. Therefore, the time t0 may be the time point representing the first 10 picoseconds before the test pulse arrives, and the time t15 may represent the time point coincident with the rising edge of the test pulse, and the time t50 may be the time point representing 10 picoseconds after the falling edge of the test pulse.

A method of depicting a graphic such as the solid line block of Fig. 8 is explained below. The intensity signal for each position and time (xi, yi, ti) has been recorded in the above description and may be referred to herein as a space-time volume. According to this embodiment, a two-dimensional color space is used, such as black and white or grayscale, and other colors may be used. Then, at each space-time volume, when the tester indicates that the transistor is a test failure, the pixel value is recorded as 1, representing, for example, white. Conversely, when the tester indicates that the transistor is a test pass, the intensity value is recorded as 0, representing, for example, black. Because the data is a cumulative value for calculating pixel values in all selected regions of the target, if the device has a 50% failure rate, the pixels will be rendered The average background color, such as gray. However, if the device has a failure rate of more than 50%, its pixels will appear light or white. If the device has a failure rate of less than 50%, the pixels will appear dark. Repeatedly scans the results to accumulate or average the results of the repeated test procedure. The charge injection of the display produces an image with a gray background color, with white pixels representing the most likely device to fail the test and black pixels representing the most likely test pass. Device. The results are shown in the dotted line box above Figure 8.

The data used to generate the graphics in the dotted box above Figure 8 can also be used to generate other graphics to show the relationship between the failure rate and time for any given pixel, as shown in Figures 8A and 8B. In particular, the user can select any pixel (device) and generate the pattern using the previously stored and corrected selected pixel intensity data, since the correction is changed to 1 or 0 based on the pass/fail test data. Therefore, the modified intensity value is related to the failure rate. As also shown in Fig. 8B, when the failure rate of the transistor at a particular pixel position changes, the pixel color changes. This correlation can be a graphical representation of the intensity signal versus time, indicating the relationship between failure rate and time.

As shown in Figure 8A, the white or white pixels in the graph represent a failure rate above 50%, while the dark dots represent a failure rate below 50%. However, the graph of FIG. 8A not only shows the failure rate, but also shows the timing relationship of the laser pulse with respect to the test clock signal, and the effect on the failure rate. For example, Figure 8A1 indicates that when the arrival time of the laser pulse coincides with the rising edge of the test pulse, the corresponding transistor typically does not pass the test. 8A2 indicates that when the arrival time of the laser pulse coincides with the falling edge of the test pulse, the corresponding transistor usually does not pass the test. Such information can be used to profile the failure mechanism of individual transistors.

In an embodiment of the present invention, a letter can be extracted from the recorded test data. A digital timing diagram to systematically progress with the transistor switching event. These timing diagrams can be constructed from a repeated test cycle of each pixel dwell time or a virtual random test of each pixel dwell time. The pulse action time in each test cycle, as well as the LADA signal intensity value obtained at each transistor position, must be known at construction time and can be obtained using short laser pulses. In an embodiment of the present invention, spatially locating or staying the incident laser beam on the target circuit characteristic, and checking the arrival time of the laser pulse relative to the test clock signal or the loop length trigger signal by time, and The arrival time of each laser pulse is recorded, and the intensity and signal of the obtained failure rate characteristic are recorded to obtain synchronized single-photon or two-photon pulse LADA information, and individual timing charts are established based on the information.

Please refer to Figure 9. The figure shows an embodiment of a test system 900. The system can work with traditional testers such as the Automated Test Equipment (ATE) 905. The tester can transmit the excitation vector signal and the clock signal to a device under test (DUT), i.e., microchip 910, and receive the output signal of the microchip 910. The applied power, ground, input signals, and resulting transmissions are all indicated by double arrows 924. The tester is arranged to repeat a series of test procedures and compare the results received from the DUT with the expected output and provide the pass or fail data of the test to a location pass/fail data via channel 926. A memory (accumulator) 960, which may be any standard memory or memory device. The tester is further coupled to a clock and pulse controller 920 that controls a charge carrier 930 via signal path 923. The injector directs the incident laser pulse 928 to a small area or space of the DUT 940 to interact with the DUT. The injector 930 also collects laser pulses reflected from the DUT. The position at which each of the laser pulses are concentrated can be regarded as a single pixel, and the light reflected by the position corresponds to one or a group of pixels on a display 970.

A transistor that undergoes a test that undergoes a signal conversion from low to high or high to low, which is injected at the same time as the resulting hole is received by absorbing photons from the laser pulse, which may change the conversion. Timing characteristics. There are already some methods for observing the switching phenomenon. In one embodiment of the invention, the tester is controlled using a signal 225 that is configured to repeat a selected test procedure to exhibit 50% pass and 50 without laser incident. % does not pass the test rate. For the sake of convenience, a simple example is given. Start with a low clock frequency so that the device can pass all test vector tests in the test program. The frequency is then doubled each time and the test vector is provided to the DUT until most of the tests fail. Then lower the clock frequency and continue testing until the failure rate reaches half. Other variables used to generate 50-50 test results include changing the voltage. All of the above are prior art techniques.

If the balance point is known, weighting the pass and fail values, a simple accumulator 960 can produce a neutralized value. In one embodiment of the invention, scanning only the surface of the DUT can in fact cause photons to interact with the transistor in the transition from low to high. The result is a change in the ratio of test pass and no pass, and the result can be presented in a chart or other form on device 970, such as a monitor. As long as a clock signal is supplied from the tester to the program control, the injector can be synchronized with the test program. In this way, it is possible to inject photons at the same point or the same majority of the repeated areas to be tested, at the time of coincidence, delay or earlyness with the test procedure. This is shown by the double arrow 922 in the figure.

During device operation, the ATE tester repeats a test procedure that includes a test program and architecture that has been set to achieve the 50% pass rate required. The data from the test is accumulated and averaged to produce an image of a gray field of view. Not in the test program At the same time, when the incident light passes through the various positions of the DUT, some positions will deviate from the 50% pass rate, resulting in a whiter or darker image point than the original gray mist.

When measuring, the device is analyzed or displayed to identify the flash point within the graphic range and the time point or position in the test program. This position is related to the transistor in a particular state. In this state, the test program is exciting a transition of the signal value at which the injector is actively injecting the laser pulse so that the energy absorbed by the single or two photons is sufficient to generate a charge carrier pair. The method can further include the step of tracking the propagation of the state transition behavior due to charge carrier injection within the layout of the circuit.

In addition, the failure rate can also be converted into a spectrum for tracking or display. For example, Fourier transforms can be used to process failure rate data collected at various locations within the target area to produce a spectrum for each location. The peak of the Fourier transform will be correlated with the failure rate at that particular location.

The method of recording and analyzing data using the above apparatus may include: constructing a tester for executing the DUT test program such that the ratio of the test failure is 50 when the transistor in the DUT is not applied with a laser pulse. %; applying the test procedure to the DUT, respectively, while illuminating the selected focus position of the DUT with a laser pulse that is built to have sufficient energy to inject charge carriers at precise time points in the test procedure Recording the test result of the tester for each focus position each time the test step is repeated; determining the position of the test result substantially deviating from the failure rate of 50% and the program point; generating a failure indicating that the failure rate deviates from the position of 50% A time series of rates versus time progression. In detecting a device that fails the test, the method of the present invention can be coupled using a transistor layout generated from the resulting location and the design network of the device under test. In addition, a pattern can also be generated to show the change of signal propagation due to charge injection. position. In an embodiment of the invention, the arrival time of the laser pulse can be scanned with time relative to the tester clock signal or the loop length trigger signal to obtain a LADA image, thereby obtaining a multi-structure LADA in a defined target region. Behavioral information, whereby most of the timing diagrams (ie, timing diagrams) can be obtained simultaneously.

In an embodiment of the present invention, an integrated package software, such as the NEXS (Navigation Exchange Server) software of Applicant Company, can be used to generate a synchronized single photon or two-photon pulse LADA for the extrapolated physical area on the DUT. Basic timing diagram. This tool provides automated time-space verification of LADA active parts using servo-controlled CAD and schematic management tools. This application can also process most targets at the same time, when individual LADA active sites or spatially defined LADA active sites are being tested and evaluated.

In an embodiment of the invention, the circuit description can be improved using the techniques described above to display a logical path connecting the LADA locations contained within the race condition. This circuit description can be used to create a planar circuit diagram for displaying the logic route in a common network format (eg, SPICE or Verilog). For example, corresponding input blocks and output blocks in a particular circuit feature can be labeled accordingly.

In an embodiment of the invention, it may be desirable to take out circuit diagram information for the subunits to direct the laser to stay in position between the associated transistors in the logic path. The results produced are beneficial for circuit debugging at the transistor level.

In an embodiment of the invention, the kit software can be applied to the analysis of two-photon induced disturbance sites. The analysis results can be applied to the separation of positive and negative transistors, and the CAD assembly for circuit analysis and image alignment.

The so-called single event upset (SEU) refers to a state change because charged particles such as ions, electrons, etc., or electromagnetic radiation strikes a microelectronic device, such as a microprocessor, semiconductor memory, or power transistor. Generated by the node. This change in state is due to ionization in or near an important node of a logic element (eg, a "bit" of a memory), resulting in a free free charge. As a result of this blow, an error occurs in the output or operation of the device, which is called SEU or soft error. Another aspect of the invention is to obtain an absorption induced disturbance point. The feedback circuit of the flip-flop is more prone to interference than the statistical combination logic gate. In a scan type test, the flip-flop operates in almost every test cycle. A single laser induced disturbance in any test cycle caused the test to fail. However, this test failure is not the same as the failure of the competitive conditions designed for LADA. Most functional flip-flops do not operate in every cycle during functional testing or operation. In this test, the disturbance point has the opportunity to subsequently generate a disturbance at each laser pulse. Therefore, at a chance of 1000+ per test, the initial low perturbation probability will become a cumulative probability of almost 100% (assuming a 100MHz laser repetition rate). As a result, a sharp but non-destructive disturbance point is observed. It should be noted here that the disturbance point can be easily removed by modifying the test procedure. The methodology disclosed herein also provides a method for measuring non-destructive two-photon absorption-induced disturbance points using an LADA-based mechanism. This methodology provides a completely new approach to the generation and evaluation of single event disturbances.

In the single photon or two-photon LADA test described herein, the laser pulse is designed to translate the switching time of the pulsed transistor. This is done by delaying or prematurely the laser pulse relative to the time of the transistor switch, causing it to pass or fail the test. This test is usually to detect if the transistor is switched at the set time. However, it should also be noted that the pulse needs to have sufficient energy to change or disturb the state of the individual storage unit. For example, such as If the flip-flop is set to store "0", a pulse with sufficient peak energy is reached, and the flip-flop can be changed to the opposite state, that is, "1" without damaging the device. This inversion can be detected by the ATE and reports the results measured at each location of the laser pulse illumination. Therefore, as long as the technique disclosed in the present invention is used, particularly using the 2pLADA technology, a memory element in which a single event disturbance may occur can be easily detected.

In the LADA test described above, a series of laser pulses will include a pulse at a point in time that will cause the transistor to pass or not pass the test. Other pulses in the series have no effect on this test. Conversely, in a single event perturbation test, each pulse in the series can cause the memory cell to reverse to its opposite state. Therefore, using a series of short pulses to have high peak power is an effective method for measuring single event perturbations. In a test cycle, the greater the number of pulses, the higher the probability of a cell reversal state.

In some embodiments of the invention, and in the pulsed laser source described above, a system using an MLL type is used. However, in some embodiments of the invention, a "non-oscillator" type laser pulse system can be used as long as the resulting pulse width and jitter characteristics allow for sufficient time resolution to plot the waveform. Not only that, but an example of such a non-oscillator type can be synchronized by directly triggering the tester's trigger signal, or a pulse triggered by the controlled delay trigger signal from the tester. Either way, a femtosecond laser pulse can be used as a better trigger for effectively generating a 2pLADA. Although picosecond pulses can also be used in the present invention, they have the disadvantage of reducing the two-photon absorption (TPA) efficiency (i.e., SNR) and increasing intrusion. The wavelength range taken by the 2pLADA should be from 1250 nm to 1550 nm. Among them, 1250nm provides the best results in terms of efficiency and resolution. Conversely, nanosecond to femtosecond laser pulses can be applied to single-photon LADA, but their laser pulse period and time resolution values, spatial resolution should be considered. The trade-off between value, intrusiveness, and peak power delivered. The single-photon LADA can be used at a wavelength of 1064 nm.

It must be noted that the methods and techniques described above are not limited in nature to any particular device and can be achieved in any suitable combination of components. In addition, various aspects of the universal device are also applicable to the invention described. It is also possible to use a special device to perform the above-described inventive method steps, and to obtain more advantages.

The foregoing is a description of the exemplary embodiments of the invention, However, different variations may be made or employed by those skilled in the art, and such structures and methods may be modified upon understanding the operations described and illustrated in the specification and the discussion of the operation. However, it does not depart from the scope defined by the scope of the invention.

210‧‧‧Device under test

215‧‧‧Tester

223‧‧‧Laser pulse sequence

224‧‧‧Laser pulse sequence

225‧‧‧pulse laser source

227‧‧‧Laser pulse sequence

229‧‧‧Laser pulse sequence

230‧‧‧ polarizer

235‧‧‧ polarizer

240‧‧‧ Objective lens

245‧‧‧ signal

250‧‧‧ computer

255‧‧‧Fixed pulsed laser source

260‧‧‧Sequence Control Electronic Components

265‧‧‧Sequence Control Electronic Components

275‧‧‧ phase adjuster

Claims (20)

A laser assisted device modification (LADA) system operable to interface with a test equipment (TE) for detecting an integrated circuit device (DUT) to be tested, and comprising: timing control electronics for receiving a a clock signal, the timing control electronics can generate a synchronization signal for synchronizing the laser pulse to the clock signal; a pulsed laser source for generating the laser pulse according to the synchronization signal, from the optical device The pulsed laser source receives the laser pulse and directs the laser pulse to a desired position on the device under test (DUT); a controller is configured to operate the timing control electronic component to The time at which the pulse reaches the transistor in the DUT is set to a time synchronized with the clock time, and the laser pulse can be delayed or advanced relative to the clock signal to thereby change the transistor to be applied to the TE An electrical response of the test signal of the DUT; a single pixel sensor for detecting a laser pulse reflected from the DUT and generating a corresponding intensity signal; and wherein the controller is configured to detect the changed signal Electricity Electrically reaction member receiving the corresponding intensity signal, and utilizing the electrical response and the intensity signal to generate a pattern representing a selected region on the DUT, the relationship between electric and reaction time. A system as claimed in claim 1, wherein the graph representing the electrical response versus time corresponds to a synchronization relationship between the laser pulses and the clock signal. A system as claimed in claim 1, wherein the graph representing the electrical response versus time comprises a graph representing a relationship between a test failure rate and an arrival time of the laser pulse relative to the test signal of the TE. A system as claimed in claim 1, wherein the pulse rate of the laser pulse is set to a multiple of the clock signal. The system of claim 1, wherein the optical device directs the laser pulse to the desired position by directing a sequence of laser pulses to respective desired positions to illuminate a portion of the DUT Select the area and reach it. The system of claim 1, wherein the pulsed laser source generates a laser pulse beam, and wherein the optical device guides the laser beam by sequentially guiding the laser beam to each desired position. Pulse to the desired position. The system of claim 1, wherein the controller uses the electrical response and intensity signal to detect a test failure or a result of passing at each desired position, and when the test fails, the intensity signal is set. For the first color, when the test pass is detected, the intensity signal is set to the second color. The system of claim 7, wherein the first color is one of black or white, and the second color is the other of black or white. The system of claim 1, wherein the controller further provides a Fourier transform to the graph of the electrical response versus time. The system of claim 1, further comprising a display, wherein the controller displays an image of a target area on the display, wherein a position where the transistor failure rate is higher than 50% is displayed as a color, and the transistor failure rate is low. Displayed as a second color at 50%; the controller further displays a graph of the electrical response versus time for each selected location on the display, the location being at a location where the transistor failure rate is greater than 50% and the transistor fails The rate is selected from a position below 50%. A method for detecting a device to be tested (DUT) coupled to a test device (TE) using a laser assisted device modification (LADA) technique, the test device generating a clock signal and a test signal; the method comprising Generating a laser beam comprising a sequence of laser pulses; sequentially positioning the laser beam at each pixel location of the target region on the DUT, and performing a test cycle for each pixel location, Wherein the test cycle includes scanning the arrival time of the laser pulse with respect to the clock signal by time, and the ATE provides the test signal to the DUT, and for each space-time volume, records the test pass/fail result of the DUT, The space-time volume is defined as a combination of the space coordinates on the DUT and the arrival time of a laser pulse; and, for most test results, where the failure rate deviates by 50%, a graph of failure rate versus time is generated. The method of claim 11, wherein the laser pulse is constructed to have a wavelength and period required to produce two-photon absorption within the DUT. The method of claim 11, wherein the laser pulse is constructed to have a wavelength and period required to produce single photon absorption within the DUT. The method of claim 11, wherein the step of recording the pass/fail result on the DUT is further included in the first color when the DUT passes the test; and when the DUT test fails, the second color is recorded. A step of. The method of claim 14, further comprising the step of generating an image on the display using the first and second colors. The method of claim 14, wherein the failure rate is plotted against time The step of selecting a space coordinate on the DUT and drawing a graphic of the first and second colors with respect to time. The method of claim 16, wherein the first color is one of black or white, and the second color is another one of black or white. The system of claim 11, wherein the pixel location comprises a flip flop, and wherein the sequence of the laser pulses is configured to generate a single perturbation event at the flip flop. The method of claim 11, further comprising displaying the test pass/fail result as a pixel position where the failure rate deviates by 50%, associated with a transistor layout of a network map. A method for detecting an integrated circuit device (DUT) to be tested coupled to a test device (TE) using a laser assisted device modification (LADA) technique, the test device generating a clock signal and a test signal; The method includes: constructing the TE to provide a test program to the DUT, so that the ratio of the pass-through test is 50% when the transistor in the DUT is not applied with a laser pulse; and the test program is repeatedly provided to the DUT while providing a laser pulse to a selected focus position on the DUT, the laser pulse being set to have sufficient energy to inject a charge pair at a precise point in the test procedure; recording each focus position at each repeat test The test result of the program; the position where the test result deviates substantially from the failure rate of 50%; and the position where the failure rate deviates substantially from the failure rate of 50%, a sequence of failure rate with respect to the time course is generated.